Transparent ceramics only exhibit the highest transparency when residual porosity and process-induced cracking are minimized. Our recent studies  showed that porosity was one important factor limiting the transparency of additively manufactured spinel samples by laser direct deposition. Laser processing conditions, particularly laser power and powder flow rate during laser deposition process, were found to have significant effects on the porosity reduction. Yan et al.  found that ultrasonic vibration contributed to reduction in porosity for (nontransparent) alumina-zirconia eutectic ceramics. It was attributed to acoustic flow of the melt aiding the natural buoyant effect of gas bubbles, allowing more to escape before solidification. Traditional sintering-based methods used sintering dopants to reduce porosity and promote densification of transparent spinel ceramics. LiF is commonly used as a sintering aid for traditionally manufactured spinel ceramics to reach the highest transparency . On the other hand, doping with rare earth ions including Dy3+ and Tb3+ is useful to increase certain luminescence bands in spinel ceramics, which may ultimately be used as emitting medium and optical radiation converters . However, very few studies have been done to investigate the doping effects on residual porosity and transparency of additively manufactured transparent ceramics.
On the other hand, dopants showed positive effects in controlling crack formation during laser direct deposition of nontransparent ceramics. Niu et al.  showed that second phase doping of alumina with yttrium aluminum garnet (YAG) and zirconia both significantly reduced cracking in laser direct deposited ceramics. Single-bead walls prepared with alumina/YAG had notably reduced cracking, and cracks were completely eliminated in alumina/zirconia parts at the eutectic ratio as a result of significant microstructural refinement. In another study, Niu et al.  showed the crack suppression effect of TiO2 dopant on Al2O3/Al2TiO5 composites fabricated by laser direct deposition. A mismatch in the coefficient of thermal expansion (CTE) resulted in compressive residual stresses in the resultant Al2TiO5 matrix. It promoted crack deflection and crack pinning, both of which were conducive to consumption of crack propagation energy.
The reduction in crack formation within laser direct deposited ceramics is closely related to the obtained mechanical properties after the introduction of dopants. The addition of zirconia to alumina had great effects on both the microstructure and mechanical properties of deposited ceramics [31,32]. Microhardness increased due to grain refinement and precipitation hardening. Fracture toughness monotonously increased as a result of transformation toughening and cracking mechanisms including crack bridging, branching and deflection. Similarly, Wu et al.  found that TiO2 doping in Al2O3/Al2TiO5 composites resulted in a maximum of 30% improvement in fracture toughness over pure alumina at a relatively low dopant percentage. Liu et al.  also showed improved fracture toughness for eutectic alumina/zirconia ceramics prepared by laser direct deposition. It was proposed to be caused by alternating residual stress fields formed during cooling due to a mismatch of thermal expansion coefficients, which resulted in crack bridging and deflection during crack propagation. However, it is not clear how dopants will affect laser direct deposited transparent ceramics, in particular crack formation.
Prior to fabrication, powder agglomerates were broken up by passing through a No. 325 mesh sieve with 44 µm opening size. Powder flowability was significantly influenced by moisture content of spinel powders. Hence, the powder was heated in air to 200 C for at least eight hours. All powders were also kept in an oven at 200 C to prevent water adsorption prior to deposition. This allowed consistent powder flow throughout experiments. Alumina substrates with dimensions 108 mm 53 mm 4 mm were used due to thermal expansion compatibility with deposited spinel ceramics. The use of alumina substrates also helped minimize substrate contamination of deposition, which would reduce spinel part transparency.
During the fabrication process, a continuous cylinder sample was printed perpendicular to the substrate surface. Ceramic powder was continuously fed into the melt pool as the CO2 laser displaced in the vertical direction. A laser spot size of 5 mm was used in this study. High laser intensities and temperatures fully melted the delivered powder and hence formed the deposited samples. Immediately after deposition, powder flow was shut off to allow the cylinders to cool in ambient air, without the influence of the conveying gas or powder flow. It is worth noting that the vertical printing strategy was implemented for this study to simply the fabrication process as inspired by the Verneuil method for flame fusion of gemstones . In addition, as suggested by previous studies [47,48], understanding the vertical build approach would facilitate fabrication of freestanding and lattice structures or even internal complex features without support materials. This would be especially beneficial for AM of ceramics. Due to high deposition temperatures and melting point of ceramic materials, support materials would be very difficult to remove.
Porosity, crack formation, and grain size were all characterized through image analysis of obtained cross-sectional images. FIJI image analysis software  was used to binarize and threshold images to only show pores and cracks, respectively. The built-in particle analysis function was used to determine the porosity area percentage. Crack formation was characterized by total crack length, average crack length, and crack density. Crack length was first measured by a ridge detection plugin . The total crack length was calculated based on the summation of the length of all cracks present on the cross-sectional surface, while average crack length was the total crack length divided by the number of cracks. Crack density was further obtained through dividing the total crack length by the cross-sectional area of the sample analyzed. Calculated equivalent circular diameters  were used to characterize grain size and were obtained from optical microscope images: the images were first post-processed in the FIJI image analysis software to determine the area of each individual grain; based on these data, average grain area was then calculated and converted to equivalent circular diameter.
Another prevalent defect within nearly all printed samples was cracking as noticed in Figure 1B and Figure 3. It was more obviously observed in the cross-sectioned and polished pure spinel samples in Figure 4. A radial crack pattern was observed, indicating the presence of circumferential thermal stress distribution during laser direct deposition of cylindrical structures. Residual cracks scattered light transmission and limited the transparency of certain regions.
On the other hand, as highlighted in Figure 5, the addition of silica dopants lowered crack formation within the printed spinel samples. The cracks were even eliminated at one 10 wt.% silica doped sample, showing the potentials of doping in significantly reducing cracks during laser direct deposition of transparent spinel ceramics. However, the obtained transparency severely deteriorated at increasing doping level. Compared to the pure spinel samples in Figure 4, the addition of merely 0.5 wt.% silica dopants lowered the obtained transparency shown in Figure 5A. Further increase of dopants drastically reduced the optical transmission, indicating the need of minimizing the doping level. Although the introduction of dopants during laser direct deposition clearly lowered the crack formation as shown in Figure 5, it will be necessary to understand how to more efficiently control crack formation with a minimal doping level so that high purity transparent spinel ceramics can be fabricated. Thus, detailed investigations on laser direct deposited spinel ceramics were carried out below to further characterize the effects of silica dopants on the microstructure, composition, and mechanical properties. In particular, this study examined residual porosity and secondary phase, the presence of which will increase light scattering due to distinctly different refractive indices compared to that of spinel [18,19,53].
Comparison of cracking (highlighted) and transparency for polished spinel samples printed with silica dopant percentages of (A) 0.5 wt.%, (B) 3 wt.%, (C) 5 wt.%, and (D) 10 wt.%.
Backscattered electron SEM images of typical polished spinel samples printed at a laser power of 580 W with silica doping percentages of (A) 0 wt.%, (B) 5 wt.%, and (C) 10 wt.%. The inset images show close-up views of the obtained microstructure with presence of micro-cracks for pure spinel samples and silica phases after the addition of silica dopants.
Results for processing induced cracking: (A) shows total crack length and average crack length, and (B) shows crack density measured for spinel samples printed at a laser power of 580 W with varying silica doping percentages.
In accordance with the observations in Figure 5, silica doping reduced crack formation, particularly the total crack length in Figure 14A and the crack density in Figure 14B, showing its feasibility in crack control for laser direct deposited spinel ceramics. The addition of merely 0.5 wt.% silica contents drastically decreased both total crack length and crack density. Increased silica dopants reduced average total crack length by up to 79% and average crack density by up to 71%. However, further increasing silica dopants in general only exhibited a moderate reduction trend in cracking. It is also worth noting that a similar trend in crack reduction was also observed when using TiO2 as dopants during laser direct deposition of nontransparent alumina ceramics . The consistent findings suggested that a low doping level may be sufficient to achieve an efficient crack reduction control. This is particularly necessary in consideration of the deteriorating effect of high doping percentages observed in Figure 5 on optical transmission. It will thus be necessary to study doping mechanisms in crack control so that a minimal level of dopants can be used in order to efficiently minimize crack formation within the printed transparent samples. It is worth noting that the variation in the reduction of total crack length and crack density in Figure 14 could be attributed to the doping variation due to silica vaporization during deposition as well as increased porosity as shown above. 153554b96e